I need a PCB designed, that takes a straightforward
"system on a chip" type intel 386 compatible processor,
and also place some high speed A/D IC and glue logic together.
The A/D can talk to the CPU via PCI or any other extremely highspeed
method. data transfer has to be via DMA, into system RAM.
So using existing schematics, etc of systems with PCI bus, is
probably the way to do it.
The A/D will be in the 40 Million Samples / Second, this could
be one that does 40Ms/s or 2 that do 20Ms/s, etc.
A open source schematic of high speed A/D in operation is here:
[url removed, login to view]
There also needs to be a standard set of digital outputs,
and this can be integrated with the control logic that
would have to send out data to configure the A/D's.
So with the A/D unit(s) there also needs to be some highspeed
digital I/O, for not only configuring the A/D chips
(i.e. they need to have setting for sample rate, polar/bi-polar,
type of sampling, scaling, etc,) but also to have some digital
I/O to control some other future logic. This digital I/O would
of course be high speed as well, as it would essentially be
hooked close to the PCI bus, and its speed.
Since this project at this stage is a proof of concept, and what
is created will not be directly marketed, any opensource,schematics can be used for this with out any legal issues,
as this device will never be sold, its strictly for internal use and
proof of a concept, that later will involve a dedicated device
The PCB has to be a fully functioning computer that would
boot linux from a IDE drive (bus) as part of it, or if one prefers,
it can boot from USB drive. The PCB must also have a
ehternet 10/100 port. No monitor, keyboard, mouse, sound, etc
would be needed, the plan would be to boot it with a
copy of linux that would have drivers for the ethernet card,
detect it, and then allow access via network. The only other
device on the PCB that needs to get access is the A/D unit(s)
and the Digital I/O via PCI bus.
Also the networking could be achieved by just having
a slot on the PBC to the PC and put in a standard ethernet card
that way, if that is preferable to schematic'ing one on to the PCB.
The linux will have to be able to control the a/d units,
and get , for example, 40M s/s data over DMA and into RAM.
The PCB can have either DDR type ram sockets on it, or DDR ram
soldered right to it, since I would like to perhaps put
2 or more GB of ram on the PCB, perhaps it should just use a
typical memory controller logic and DDR ram slots, and allow expansion.
I am not sure what bios is available to use, but i have to assume
generic PCB board intel 386 compatable computer designed schematics
are out there for free, or provided by chip manufactures, like
AMD and intel (and VIA, etc) so one can work with their products.
the only real "new" work on this project is interfacing a
A/D unit(s) to the PCI bus and getting the data to the RAM of the
computer. As well as controlling Digital inputs/outputs.
It would probably be preferable to have a daughter card with the
A/D unit(s) on it, so the function of the computer
can be diagnosed independant of the A/D. So that if the
A/D circuit doesnt work, or is buggy, only it would have to be
re-etched, but the main computer part would be be isolated,
and not have to be fab'd.
Creating the computer part with a daughter card interface,
a IDE bus and a ethernet and PCI bus should be a standard
available schematic, and layout files out there,
again the real work comes in making a A/D unit to attach to it.
The quote would involve the following being provided:
1) final schematic and layout file for PCB to be etched
2) necessary documentation to back up the design, and essentaily
proof that it should work.
It would then be prefered that the designer also take delivery of
the fabricated product from a etch/solder shop, and assemble, and
install linux, check that ethernet is working and have connectivity,
as well as test for necessary data capture, in effect testing the
The quote would not include any material cost or fabrication cost.
The quote just needs to include:
2) assembly of fab'd pcb, i.e. plug in i.e. pentium 4 PRocessor into
cpu socket, plug in ram, also mount pcb board to a typical housing,
for protection and workability.
3) write a small amount of code, probably in C (there is
A/D pci card open source code in linux kernel for talking with
other manufacturers pci based A/D boards, and working with
typical PCI data interface), and thus getting the A/D captured
data into the systems RAM.
This project could be divided into 3 seperate and paid for pieces,
upon each ones completion, i.e.
1) research, getting right components (i can aid in this).
2) design schematic and PCB layout file that a fab shop accepts
3) assemble and test
It also might be appropriate that two people work on this, one that
has more board level and PCB knowledge, and another with C/linux
device coding skills.
For a very experienced person who has done all this for others,
this might be less then 2 weeks of work, it could also be 3-4 weeks,
for a lesser experienced person.
I think it would be important, to assure good sucess in this project,
to acquire known working schematics, and pcb files for a working
PCB intel compatible system (with pci) and a schematic for
a a/d unit (i included a link to one above), and
have a good feeling that this is more about glueing together known
existing working technology then re-inventing the wheel.
I choose the ($1500-$5000) large budget setting but i don't
believe it should be nearly as high as $5000, nor is it necessarily
as low as 1500$, however a person who has essentailly already
done this kind of work, could easily do it for that, given its
just pulling from past work and experience.
And remember, all part cost and PCB fabbing, is covered,
and not to be included in the quote.
This type of project is very specialized, so i will mark it as
35 days, and then repost, i am not in a hurry, so this could make
some nice part time hobby work for someone, even a college student,
9 freelancers are bidding on average $1912 for this job
Dear Sir, i'm a graduate of Electrical and Electronics Engineering. So this will be a simple job. I have protel as my tools. I also know much about embedded [url removed, login to view] just trust me for quality of work.Thanks.
I have put Electronic Engineers on this project. We are mainly interested in A/D software development. Need some clarifications, can you please give us time till July 5th 2006. Saras Engineering Services (SES)